Search papers, labs, and topics across Lattice.
2
0
6
0
MLLMs can recognize urban scenes but fail to maintain reliable navigation and goal-directed behavior over extended exploration in complex environments.
Salience Bias in LLMs reveals that models often ignore commonsense reasoning in favor of misleading explicit cues, with lightweight prompting showing promise in addressing this issue.
